In multi-layer and multi-domain networks using Generalized Multi-Protocol Label Switching (GMPLS), multiple types of nodes having different switching capabilities and rates can be divided into different domains or layers, and these domains may be router domains, Optical Transport Network (OTN) domains or Packet Transport Network (PTN) domains and the like. In a multi-domain network divided by an Autonomous System (AS), an end-to-end path transversing multiple domains can be computed using a Routing Controller (RC) or a Path Computation Element.
The Optical Internetworking Forum (OIF) is a standard organization specialized in studying optical internetworking. In order to implement intercommunication between multi-domain networks, a standard of “E-NNI OSPFv2-based Routing −2.0 Implementation Agreement” has been formulated and a multi-domain intercommunication path computation method using an RC and based on the Domain to Domain Routing Protocol (DDRP) has been proposed. In such a method, intra-domain topology of each domain is abstracted and represented as abstract links, inter-domain topology information of multiple domain is flooded through the DDRP, which includes information regarding inter-domain links, intra-domain abstract links, and boundary nodes of each domain, as shown in FIG. 1. Based on the inter-domain topology, computation of inter-domain routing is performed by an RC of a first domain, then based on computed inter-domain paths (including inter-domain links, domain boundary nodes and intra-domain abstract links), specific intra-domain routing in each domain is computed at entering boundary nodes of each domain, and signaling is routed according to the computed path, as shown in FIG. 2.
The Internet Engineering Task Force (IETF) has proposed multi-domain path computation using a PCE. Specifically, RFC5441 proposes a Backward-Recursive PCE-Based Computation (BRPC) algorithm. This algorithm depends on communication between collaborative PCEs. A Path Computation Client (PCC) transmits, to PCEs within its respective domains, a path computation request PCReg message. The PCReg message is forwarded among PCEs of respective domains until the request reaches a PCE of a domain where a destination node taking charging of computation of a Label Switching Path (LSP) is located. The PCE in the destination domain establishes a
Virtual Shortest Path Tree (VSPT) consisting of potential paths reaching the destination node, and transmits the path tree, via a path computation response PCRep message, to a preceding PCE. Then the VSPT is added to respective PCEs in sequence and transmitted backwards until reaching a PCE of a domain where a source node is located, the PCE selects, using the VSPT, an optimum end-to-end path and transmits the path to the PCC, as shown in FIG. 3. The method shown in FIG. 3 requires respective domains involving in the computation based on BRPC algorithm to support PCEs and requires all PCEs in respective domains to support an RFC5088 flooding automatic discovery function so that all the PCEs in respective domains are known to its neighbors. Certainly, the RFC5152 from the IETF further proposes a per-domain routing computation method with its basic principle similar to the RC method from OIF.
At present, in practical applications, there are three scenarios: 1) if respective domains in a domain sequence don't support PCEs, the RC computation method from OIF is adopted; 2) if respective domains in a domain sequence all support PCEs, the PCE BRPC computation method from IETF is adopted; and 3) if there are some domains supporting PCEs and some domains not supporting PCEs, an intercommunication computation is required to be performed.
As to the scenario 3), typical methods currently used include for example a method described in Chinese patent application published on Oct. 29, 2008 with the publication No. CN101296178A, i.e., a segmented computation method based on the per-domain routing computation and the BRPC respectively is adopted, it is determined by a boundary node of a preceding domain whether a next domain has PCEs, if yes, computation is performed according to a first path computation method (BRPC algorithm for PCEs); otherwise, computation is performed (as shown in an autonomous system 2 in FIG. 4) according to a second path computation method (per-domain routing computation algorithm). Finally, the routing is computed using the first path computation method or the second path computation method in sequence in respective domains, and lastly establishment of signaling is completed according to the computed routing. This method though partially solves the problem of intercommunication computation, there are still following defects: for a domain where the BRPC algorithm for PCEs is applied, it is still required that consecutive domains are all deployed PCEs, if there is a domain deployed without a PCE between two domains deployed with a PCE, the BRPC algorithm cannot be used, instead only the per-domain path computation method can be used, thus this may greatly lower the optimality of path computation and result in a poor efficiency of the path computation.